Exceptionally high decarboxylation rate of a primary aliphatic acyloxy radical determined by radical product yield analysis and quantitative 1H-CIDNP spectroscopy

Article abstract of DOI:10.1002/poc.1836

Symmetrical (RCO₂CO₂R; R=XCH₂CH ₂) and asymmetrical (RCO₂CO₂R′; R=C ₉H₁₉CH₂CH₂, R′=CH₃ or m-ClC₆H₄) primary diacyl peroxides were thermally decomposed under different conditions to analyze the decarboxylation rates of the thermally generated acyloxy radicals. Quantitative models of the geminate product yields, and qualitative and quantitative ¹H-CIDNP spectroscopy were used to obtain the decarboxylation rate estimates. Results reported here suggest that, unlike short chain acyloxy radicals such as propanoyloxyl, long chain acyloxy radicals possess the highest decarboxylation rates of all known acyloxy radicals, estimated at (0.5-1.5)× 10 ¹²s^-1 between 80 and 140°C. Given the nature of the dissociative state of acyloxy radicals, such rates appear to be the result of destabilization of the former by the steric bulk of the long chain substituents. Additionally, the rate of this order of magnitude suggests a nearly concerted decarboxylation of primary diacyl peroxides. Copyright

Full text of DOI:10.1002/poc.1836

Research Article
Received: 2 April 2010,
Revised: 8 December 2010,
Accepted: 20 December 2010,
Published online in Wiley Online Library: 2 March 2011
(wileyonlinelibrary.com) DOI 10.1002/poc.1836
Exceptionally high decarboxylation rate of a
primary aliphatic acyloxy radical determined
by radical product yield analysis and
1
quantitative H-CIDNP spectroscopy
Alicia Frainday, Ryan Turncliffa, Teri Foxa, Justin Sodanoa
and Lev R. Ryzhkova
*
Symmetrical (RCO2CO2R; R ¼ XCH2CH2) and asymmetrical (RCO2CO2R(; R ¼ C9H19CH2CH2, R( ¼ CH3 or m-ClC6H4)
primary diacyl peroxides were thermally decomposed under different conditions to analyze the decarboxylation
rates of the thermally generated acyloxy radicals. Quantitative models of the geminate product yields, and qualitative
1
and quantitative H-CIDNP spectroscopy were used to obtain the decarboxylation rate estimates. Results reported
here suggest that, unlike short chain acyloxy radicals such as propanoyloxyl, long chain acyloxy radicals possess the
highest decarboxylation rates of all known acyloxy radicals, estimated at (0.5–1.5) T 1012 sS1 between 80 and 140 -C.
Given the nature of the dissociative state of acyloxy radicals, such rates appear to be the result of destabilization of
the former by the steric bulk of the long chain substituents. Additionally, the rate of this order of magnitude suggests
a nearly concerted decarboxylation of primary diacyl peroxides. Copyright ß 2011 John Wiley & Sons, Ltd.
Supporting information may be found in the online version of this article.
Keywords: acyloxy radicals; CIDNP; decarboxylation rates; radical pair reactions
When generated photolytically from the corresponding
N-hydroxy-2-pyridinethione esters, both 3 and 5 give rise to
INTRODUCTION
Acyloxy radicals, 1 (RCO:2), are valuable intermediates in the Barton
PTOC ester[1] and related reactions, where they serve as versatile
precursors of ‘‘well behaved’’ alkyl radicals used in synthesis of
such complex natural products as interiorins, kadsulignans,[2] and
mitomycins.[3,4] They are also key intermediates in the thermolysis
of diacyl and related peroxides, commonly used polymerization
initiators.[5] Frequently, they are precursors used in calibration of
many of the ‘‘radical clocks’’[6,7] that probe the rates of very rapid
enzymatic processes, such as hydrocarbon oxidation by cytochrome
P-450.[8,9] They are also reactive intermediates in organic reactions
ranging from thermal or photodecomposition of diacyl per-
oxides[10–15] and other acyloxy derivatives[16] to Kolbe electrolysis
and Hundsdiecker reaction,[17] and are involved in some atmospheric
photochemistry of oxygenated fluorocarbon radicals.[18,19]
Knowledge and prediction of decarboxylation rates is essential
if one is to understand the distribution of radical-derived
products from acyloxy radical precursors fully and quantitatively.
This can be illustrated by the yet unexplained differences
in behavior of related, primary acyloxy radicals trans or cis-2-
substituted cyclopropylcarbinyl 3 (R ¼ trans-ethoxy- or trans-
tert-butoxycarbonyl[20] or cis- or trans-phenyl),[21] and 5 (R ¼
various substituted aryl).[22]
short-lived ‘‘radical clocks’’ 4 and 6, which undergo very
rapid ring-opening, k ¼ (0.7–5.0) ꢀ 1011 sꢁ1 at 25 8C. While the
pre-exponential factor for the ring opening of 6 is 13.1, as
expected from thermochemical estimates, the corresponding
values for 4 are anomalously high, 13.8–14.3. The difference in
rearrangement kinetics of 4 and 6 may depend on the stability
of the precursor acyloxy radicals. It is possible that radicals of
type 3 may decarboxylate in concert with ring-opening.[22]
This may indicate a decarboxylation rate of the ‘‘primary’’
acyloxy radicals (RCH2CO:2 or R0CH2CH2CO2: ) to be much higher
than the usually assumed (2–5) ꢀ 109 sꢁ1 at RT (vide infra).
Decarboxylation rates of some radicals are well established,
Table 1. For aroyloxy (1a, R ¼ substituted phenyl) and acetoxy (1b,
R ¼ CH3) radicals they have been measured, or estimated, by
many direct and indirect methods,[23–28] and are typically on
the order of 108 sꢁ1 at 80 8C, with the activation energy of
8–9 kcal/mol for 1a, and 109 sꢁ1 at 80 8C with the activation
*
Correspondence to: L. R. Ryzhkov, Department of Chemistry, Towson Univer-
sity, Towson, MD 21252-0001, USA.
E-mail: lryzhkov@towson.edu
a
A. Fraind, R. Turncliff, T. Fox, J. Sodano, L. R. Ryzhkov
Department of Chemistry, Towson University, Towson, Maryland 21252-0001,
USA
y
Present address: Chemistry Department, The Johns Hopkins University, 138
Remsen Hall, 3400 N. Charles St., Baltimore, Maryland 21218, USA.
J. Phys. Org. Chem. 2011, 24 809–820
Copyright ß 2011 John Wiley & Sons, Ltd.

A. FRAIND ET AL.
Table 1. Lifetime estimates of acyloxy radicals
Radical
Lifetime (ps)
Radical
Lifetime (ps)
Aroyloxy 1a; R ¼ substituted phenyl
Acetoxy 1b; R ¼ CH3
10 000 (25 8C)
1000 (120 8C)
560 (25 8C)
Naphthyloxycarbonyloxy
Propanoyloxy 1c; R ¼ C2H5
Ar2C(OH)CO2
33 (25 8C)
10–2 (120 8C)
1.2 (25 8C)
N/A
Arylacetoxy 1e; R ¼ PhCH2
3-Phenylpropanoyloxy 1f, R ¼ PhCH2CH2
430–218 (25 8C)
Dodecanoyloxy 1d; R ¼ n-C11H23
and related long-chain radicals
9-Methylfluorenylcarbonyloxy
55 (20 8C)
energy of approximately 7 kcal/mol for 1b. The estimates of
decarboxylation rates for primary acyloxy radicals such as 1c
(R ¼ C2H5), when the latter are generated photolytically at room
temperature, are similar. Thus, for a variety of arylacetoxy radicals
(1e, R ¼ PhCH2), these rates range from 1.5 ꢀ 109 to 4.6 ꢀ 109 sꢁ1
when estimated from the product distribution studies in
photolysis of 1-naphthylmethyl phenylacetates.[29,30] The latter
study also provides the rates for the decarboxylation of 1c and
3-phenylpropanoyloxy (1f, R ¼ PhCH2CH2) radicals (2.0 and
2.3 ꢀ 109 sꢁ1, respectively, at RT). Radicals with a single oxygen
atom between the aromatic group and the CO2 moiety, such as
the benzyloxycarbonyloxy and naphthyloxycarbonyloxy, exhibit
CHCl3. We base this work on our previous detection of the
disproportionation and combination products from short-lived
acyloxy-alkyl radical pairs, Cage II in Scheme 1, in thermolysis
of very low (10ꢁ3–10ꢁ4 M) concentrations of 2 and related
primary peroxides.[10,11] The acyloxy-alkyl radical pairs were
generated by the homolysis of the peroxide O–O bond, and
subsequent decarboxylation of one of the acyloxy radicals.
This communication extends our previous studies of 2 and
symmetrical peroxides 8a, 8b, and 8c to mixed peroxides 9a
and 9b to provide the previously unreported lifetime esti-
mates for the primary acyloxy radicals 7. These estimates were
obtained from the quantitative yield analysis of cage products
from 2, and from the analysis of the qualitative and quantitative
1H-CIDNP spectra obtained during thermal decomposition of
2 and 9a.
similar lifetimes (0.2 and 3.0 ꢀ 1010
s
ꢁ1 respectively, also at RT).[31]
The lifetimes of some of the acyloxy radicals generated by
the photolysis of peresters or peroxides are pushing the envelope
of what may be considered a ‘‘discrete’’ solution-phase inter-
mediate, even at or near room temperature. For example, the
decarboxylation rate of the thermally equilibrated 9-methyl-
This paper addresses two main issues:
A. Obtaining a firm quantitative estimate for the decarboxylation
rate of 1d. The accurate estimates for the thermal decarbox-
ylation rates of primary acyloxy radicals such as 1c and 1d are
necessary since the rate for 1d has not been previously
reported, and current CIDNP and product analysis results
provide an unacceptably wide range of values for these
rates.[36–38] For example, the wide range of the decarboxyla-
tion rate values for the propanoyloxy radical 1c is a direct
result of the typically small values of the decarboxylation
activation energy (7 ꢃ 3 kJ/mol), and correspondingly large
errors in the its estimate.[36–38] The error of this magnitude is
also due to product yields reported in Refs.[36–38] being
obtained exclusively from decomposition of very high per-
oxide concentrations (0.1–1.0 M) in reactive solvents such as
methanol and acetone. It is likely that such conditions cause
artificial variation of the product yields due to the bimolecular
side reactions between the peroxide and the escaped radicals
and/or cage products. In general, all current models of fast
radical reactions in solution are subject to variations in the
viscosity of the medium and diffusion coefficients of various
species – all temperature-dependent, with very similar acti-
vation parameters.[39–42] Unless these are corrected for, the
result will be a large errors in the estimates of the activation
fluorenylcarbonyloxy radical is reported to be 1.8 ꢀ 1010
s
ꢁ1 at RT,
a 55 ps lifetime, determined by picosecond[32,33] and femtose-
cond[34] time-resolved photolysis of the tert-butyl 9-methyl-
fluorene-9-percarboxylate, while the rate for the naphthylox-
ycarbonyloxy radical is 3.0 ꢀ 1010 sꢁ1 at RT.[31] As the matter
of fact, the vibrationally hot 9-methylfluorenylcarbonyloxy
radical decarboxylates within at most 2 ps,[34] while the room
temperature decarboxylation rate of the benzilate anion-derived
acyloxy radical (Ar2C(OH)CO2ꢂ) is estimated to be 8.0 ꢀ 1011 sꢁ1
–
a 1.2 ps lifetime, determined by femtosecond time-resolved
spectroscopy, following the photoinduced charge-transfer
activation in methylviologen-carboxylate anion radical pairs.[35]
Typically, such fast rates are attributed to the high stability of the
decarboxylation product radicals, and an excess of vibrational
energy after the decomposition of the parent peroxide.[34]
In general, the long chain primary acyloxy radical 1d,
usually generated by thermolysis of a primary symmetrical
diacyl peroxide 2 (lauroyl peroxide; RCO2CO2R; R ¼ XCH2CH2;
X ¼ n-C9H19), is often used as a simple model for primary
acyloxy radicals.[10,11,36–38] This is largely due to the ready
availability of 2 as a commercial product, Alperox, and ease
of purifying it through repeated recrystallization from methanol/
wileyonlinelibrary.com/journal/poc
Copyright ß 2011 John Wiley & Sons, Ltd.
J. Phys. Org. Chem. 2011, 24 809–820

HIGH DECARBOXYLATION RATE OF AN ACYLOXY RADICAL
Scheme 1.
energy, and therefore in the wide range of the decarboxyla-
tion rate values. In this paper, we present a comprehensive
quantitative product yield analysis for peroxide 2 that
avoids these issues.
Materials
Toluene, decane, and benzene (Aldrich) were distilled from
CaH2. High purity n-hexane (Acros) and n-pentane (Fischer) were
checked by GC-MS and used without further purification.
Hexachloroacetone (HCA, Aldrich) was vacuum distilled at
80 8C. Diphenylbutane and 2-cyclopentylethanol (Alfa Aesar,
both 97%) were used as received; ethylbenzene, vinylcyclopentane
(99%), 3-cyclopentylpropanoic acid (98%), 3-phenylpropanoic acid
(99%), and ethylcyclopentane (99%) were obtained from Aldrich
and used without further purification. m-CPBA was recrystallized
from aqueous pH 7.4 phosphate buffer.[44] The procedures for the
GC, GC/MS, and 1H-NMR analysis of peroxide decomposition
product yields were reported previously.[10,11]
B. Resolution of the inconsistencies in the decarboxylation rates
and activation energies for simple primary acyloxy radicals
such as 1c and 1f (2–4 ꢀ 109 sꢁ1 at between RT and 120 8C)[12]
and those of 1b. These estimates are of the same order of
magnitude, suggesting that net CIDNP enhancements should
be observed in thermolysis of 2 and related primary diacyl
peroxides (vide infra). However, previous CIDNP investi-
gations[43] failed to detect any such enhancements leading
to a common assumption that these and other primary
aliphatic acyloxy radicals, such as long-chain 1d, decarbox-
ylate at rates at least an order of magnitude higher than that
of 1b. Indeed, the most recent quantitative CIDNP and
product yield studies of the thermolysis of acetyl propanoyl
peroxide[36–38] appear to confirm that assumption (estimated
decarboxylation rate of 1.0–10 ꢀ 1010 sꢁ1 for 1c at 80 8C). In
fact, failure to detect net CIDNP enhancements during thermal
decomposition of 2 and other primary diacyl peroxides[28,43]
was often taken as evidence against the existence of a discrete
primary acyloxy radical intermediate such as 1c or 1d.
CIDNP experiments
1H-CIDNP spectra were recorded on both Bruker AM400-WB
and WM250 instruments, equipped with a Bruker VT 1000
variable temperature control unit, the probe temperature being
monitored with DORIC Trendicator 410A digital thermometers
with the resolution of no better than ꢃ0.05 8C. NMR tubes (5 mm)
charged with 0.1–0.2 M peroxide samples were outfitted
with holed tube caps in order to permit the escape of CO2. No
field/frequency lock was used. Most of the peroxide samples
were decomposed at 140 8C, but some at 120 8C. In all cases, the
collection of FIDs started as soon as the sample was introduced
into the hot probe. The FIDs were Fourier transformed and the
resulting spectra were examined to obtain those exhibiting
CIDNP. In all cases, single pulse FIDs were stored separately
when the progressive saturation technique was used. For a
typical steady state spectra 8K or 16K FIDs were co-added in
series of 8 or 4, respectively. Progressive saturation 1H CIDNP were
obtained according to the procedure of Barbara and Lawler.[45]
A typical procedure used a string of 908 pulses applied prior to
the acquisition of the CIDNP FID following a recovery delay with a
short t smaller than the shortest T1 of the observed cage product
signals (see text). The resultant signal intensity is dominated by
the CIDNP enhancements and is independent of individual
relaxation times.
EXPERIMENTAL SECTION
General methods
Routine 1H- and 13C-NMR spectra were recorded on JEOL 90 MHz,
Bruker WM250 (250 MHz), and AM400-WB (400 MHz) instruments.
All chemical shifts are reported in ppm (d) with respect to TMS.
IR spectra were recorded on a Perkin-Elmer Paragon or 1600
Series FTIR Spectrophotometer (at 4 cmꢁ1), with the samples
observed as either thin films or in Nujol mulls. Peroxide samples
were thermalized in the NESLAB EX-210 oil bath equipped
with proportional temperature controller. Dimethylpolysiloxane
200 fluid was used and the temperature was monitored with
the mercury thermometer readable to ꢃ0.05 8C. Samples were
thermalized in sealed glass ampoules (0.4 mm i.d., ca. 2 in. in
length).
J. Phys. Org. Chem. 2011, 24 809–820
Copyright ß 2011 John Wiley & Sons, Ltd.
wileyonlinelibrary.com/journal/poc

A. FRAIND ET AL.
Preparation of compounds
and a less pure material containing ca. 25% of the anhydride were
the same.
The procedures for the preparation of peroxylauric acid, lauroyl
peroxide (2), peroxynonanoic acid, bis-(nonanoyl) peroxide (8a),
3-cyclopentylperoxypropanoic acid, bis-(3-cyclopentylpropanoyl)
peroxide (8b), bis-(3-phenylpropanoyl) peroxide (8c), 3-phenylethyl
3-phenylpropanoate, and 3-cyclopentylethyl 3-cyclopentyl-
propanoate were reported previously.[10,11]
RESULTS
Product yields from 2, 8a–c, and 9a,b
The decomposition of 0.1 M 2, 8, and 9 at fixed temperatures
1
ranging from 80 to 89 8C was followed by H-NMR giving first
Acetyl lauroyl peroxide (9a)
order decomposition rate constants ranging from 1.0 ꢃ 0.1 ꢀ
10ꢁ4 sꢁ1 for 9b in C6D6 at 80 8C to 3.1 ꢃ 0.5 ꢀ 10ꢁ4 sꢁ1 for 2
in toluene-d8 at 89 8C. The average decomposition rate for all
peroxides was 1.6 ꢃ 0.6 ꢀ 10ꢁ4 sꢁ1 between 80 and 83 8C.
Since the decomposition of these peroxides presumably involves
a common O–O bond homolysis step, these rate constants
are identical to 1.4 ꢀ 10ꢁ4 sꢁ1 obtained previously for 2 under
identical conditions.[46–48] The sensitivity of rate to structure of
the R group, for example, primary, sec, tertiary, benzyl, etc. has
also been used as evidence for multiple bond cleavage.
Thermolysis yields of excess alkene from 2 and 8a–c and 9a,b
at initial peroxide concentration <10ꢁ3 M, that is, under the
conditions of complete oxygen scavenging of escaped radicals,
are presented in Table 2. As reported previously,[11] the yields
were defined so that complete conversion of the peroxide to a
product by a single type of reaction would be described as a
100% yield of that product. The yields were also corrected by
dividing by the fraction of the peroxide decomposing by radical
pathway only, since some of the peroxides studied here can
partly decompose by the ‘‘polar pathway,’’ Scheme 2,[47,49,50]
through the intermediacy of the rearranged carboxyinversion
product 10. The fraction of the peroxide decomposing by radical
pathway was estimated by combining the yields of all radical
products listed in Tables 2 and 3.
Complete scavenging of the radicals under these conditions
was previously confirmed for 2, and 8a–c.[10] In fact, for the very
closely related peroxides 2 and 8a, the yields of alkenes
(6.8 ꢃ 0.2% for 1-undecene from 2; 6.7 ꢃ 0.2% for 1-octene from
8a) and scavenging products (12.3 ꢃ 1.8% for n-undecylbenzene
from 2; 12.2 ꢃ 1.6% for n-octylbenzene from 8a) were identical
when determined from 0.1 M 2 and 8a in C6D6. A similar analysis
was performed for excess 1-undecene yields from mixed
peroxides 9a and 9b. As can be seen from Table 2, the yields
are higher in 9a and 9b. This is due to the longer lifetimes of the
acetyloxyl and m-chlorobenzoyloxy radicals (vide supra) resulting
Acetyl chloride (1.5 mL, 0.02 mol) was added to the solution of
peroxylauric acid in 40 mL of diethyl ether and the mixture cooled
to 0–2 8C using an ice-water bath. Pyridine (1.94 mL, 0.024 mol)
was added with vigorous magnetic stirring in 100 mL portions
every 3 min. After the complete addition of pyridine, the ice bath
was removed and stirring was continued for 30 min. Pyridinium
hydrochloride was filtered off and the ether layer was washed
twice with 5% HCl, once with water, twice with 5% NaHCO3, and
once again with water. The peroxide was recrystallized from cold
ether, 1.25 g. (30% yield). The rest of the material contained ca.
15% of both lauric acid and lauroyl anhydride and was not used.
IR (Nujol mull): 1820, 1791 cmꢁ1. 400 MHz 1H-NMR (CDCl3): d 2.45
(t, 2H), d 2.2 (s, 3H), d 1.75 (quintet, 2H), d 1.3 (broad, 16H), d 0.9
(t, 3H). All spectroscopic data for 9a and 9b was in agreement
with published values.[46]
3-Chlorobenzoyl lauroyl peroxide (9b)
Lauroyl chloride 2.18 g (0.01 mol) and m-CPBA 2.0 g (0.011 mol)
was added to 45 mL of diethyl ether in 100 mL round bottom
flask. The mixture was cooled with an ice-water bath and 0.83 g.
(0.011 mol) of pyridine added dropwise. After the addition the
reaction mixture was stirred for an additional 45 min, washed
twice with 5% HCl, twice with 5% NaHCO3, and twice with
distilled water. Yield: 1.57 g. (44%). IR (Nujol mull): 1795,
1767 cmꢁ1. 400 MHz 1H-NMR (CDCl3): d 8.0 (t, 1H), d 7.9 (doublet
of triplets, 1H), d 7.62 (doublet of triplets, 1H), d 7.45 (t, 1H), d 2.5
(t, 2H), d 1.77 (quintet, 2H), d 1.4 (broad multiplet, 2H), d 1.3
(broad, 16H), d 0.9 (t, 3H). 400 MHz 13C-NMR (CDCl3): d 169.1, 161.9
(carbonyl groups), d 135.05, 134.6, 130.1, 129.7, 127.8, 127.4
(aromatic), 11 additional peaks between d 32 and 14 ppm
(undecyl group). The product contained ca. 1–2% of 3-
chlorobenzoyl lauroyl anhydride (dH 2.61 ppm, triplet) which
proved to be impossible to remove even after repeat
recrystallization. The product yields and CIDNP from this peroxide
Table 2. Cage II disproportionation product yields from peroxides 2, 8, and 9 in n-hexane at 80 8C
Conditions
2
Alkene (E)a
6.7 ꢃ 0.2
6.0 ꢃ 0.3
7.4 ꢃ 0.2
8.0 ꢃ 0.3
8.2 ꢃ 0.2
12.9 ꢃ 0.6
Alkane (A)
2.6 ꢃ 0.1
3.5 ꢃ 0.1
2.4 ꢃ 0.1
2.9 ꢃ 0.1
ꢀ
Excess alkene (E)b
4.1 ꢃ 0.2
2.5 ꢃ 0.1
5.0 ꢃ 0.2
5.1 ꢃ 0.3
ꢀ
Ratio E/A
1.6 ꢃ 0.2
0.7 ꢃ 0.3
2.1 ꢃ 0.2
1.7 ꢃ 0.1
ꢀ
8a
8b
8c
9a
9b
ꢀ
ꢀ
ꢀ
a % yield relative to 1 mol of decomposed peroxide. At least 4–6 different samples analyzed, 1 or 2 different ratios of int. strd. to the
decomposing 1–2 ꢀ 10ꢁ4 M peroxide. These data are an average of peroxide decompositions in n-octane or n-pentane/iso-octane
mixtures or n-pentane. Errors represent one standard deviation. The thermolysis time was sufficient to decompose >98% of peroxide.
b Yield of alkene ꢁ yield of alkane.
wileyonlinelibrary.com/journal/poc
Copyright ß 2011 John Wiley & Sons, Ltd.
J. Phys. Org. Chem. 2011, 24 809–820

HIGH DECARBOXYLATION RATE OF AN ACYLOXY RADICAL
Scheme 2.
in higher disproportionation yields from Cage II. Regardless, the
radical product yields in these peroxides show that excess alkene
from the acyloxy-alkyl radical pair is observed from all these
peroxides, and it is sensitive to the lifetime of the acyloxy radical
effecting the disproportionation in Cage II, Scheme 1.
yields of 1-undecene from these peroxides are shown in Table 4.
Clearly there is no systematic variation of the ester yield with
solvent polarity. It is also possible to check if ester comes from
the thermolysis, during the GC analysis, of the carboxyinversion
product – Scheme 2. Thus, the amount of ester, methyl
undecanoate, from thermolysis of 9a was estimated by NMR
prior to the injection of the sample into 200 8C GC inlet (11.0 ꢃ
1.3%) and compared to the value obtained from the GC analysis
(9.8 ꢃ 0.4%). As was reported previously for peroxide 2,[10,11]
there appears to be no significant amount of ester formed by
the thermolysis of carboxyinversion product.
As can be seen from Table 3, the yields of purely ‘‘polar’’
products vary widely. For example, the yields of n-octanol (8.3 ꢃ
1.2%) and n-octyl acyl carbonate (10.8 ꢃ 1.6%) from 8a were
quite different compared to similar products from 2 (2.5 ꢃ 0.3%
for n-undecanol; 7.0 ꢃ 0.8% for the n-undecyl acyl carbonate). No
carboxyinversion product 10 was detected in thermolysis of 8b,
however the yield of alcohol (8.0 ꢃ 0.7) was similar to that from
8a under identical conditions. Peroxide 8b is also notable for
the high yields of the ester, 3-cyclopentylethyl 3-cyclopentyl-
propanoate, and combination product, 1,4-dicyclopentylbutane,
both about twice as high as in the case of 2, 8a, and 8c.
The yields of combination, scavenging, and oxygen-containing
products from peroxides 2, 8a–c, and 9a,b are reported in Table 3.
As expected, the yield of combination product from Cage III
decreases in the series 2, 8a–c, 9a,b since most of the cage
products come from the cage containing the successively
longer-lived acyloxy radicals. Likewise, the yield of products from
Cage II increases in this series. The yields of undecyl acetate
from 9a and the yields of undecyl m-chlorobenzoate from 9b are
about 2 times higher than the yield of undecyl lauroate ester from
2. Very little ester from the cages with decarboxylated acetyloxyl
or m-chlorobenzoyloxy radicals is formed, none detected in the
case of 9b and 1.0 ꢃ 0.1% in case of 9a. The total yield of escape
products is very similar in all these cases if low yield of methyl
chloride from 9a is ignored since a 1/1 ratio of these methyl
chloride and undecyl chloride is expected. This is justified
because the thermolysis was carried out in the open tube to avoid
pressure buildup and some of the volatile methyl chloride
undoubtedly escaped from the sample. Note that although the
yields of esters are very similar in the case of 2 and 9a, it comes
mostly from Cage II in case of decomposition of 9a, whereas in
2 equal amounts are formed from Cages II and III. The ratios of the
coupling and disproportionation pathways for these peroxides
can be determined from the ester and excess alkene yields listed
in Tables 2 and 3. It is ca. 4:1 for the 9a and 2. However, in the case
of 9b this ratio is about 1.5:1. This raises the possibility that
1-undecene in thermolysis of these peroxides, especially 9b, is
partly produced by the polar pathway, since it has been widely
observed that the yield of polar products increase in peroxides
with aromatic ring substituents.[47,49,50] In order to confirm that all
of 1-undecene comes solely from the radical pathway, 2, 9a, and
9b were thermalized in solvents of widely different polarity. The
1
Qualitative and quantitative H-CIDNP spectroscopy
A typical CIDNP[42,51,52] experiment involves thermolysis of
peroxide samples in the heated probe of the NMR spectrometer.
The qualitative features of the CIDNP spectra presented here can
be explained in reference to Scheme 1. During the thermolysis,
the radicals generated in Cages I–III are ‘‘born’’ in a singlet
state and the resultant radical pairs remain spin-correlated during
the subsequent diffusional re-encounters. As the result, only the
pairs in the singlet state react to give ‘‘cage products’’ while
the escaped radicals are correlated as triplet pairs. Since the
electrons on individual radicals are coupled to nearby nuclei
(if the latter possess a spin, which is not the case in Cage I), and
J. Phys. Org. Chem. 2011, 24 809–820
Copyright ß 2011 John Wiley & Sons, Ltd.
wileyonlinelibrary.com/journal/poc

A. FRAIND ET AL.
Table 3. Combined 1H-NMR and GC/MS of combination and other product yields from thermolysis of 0.10 M 2, 8, and 9 at 80 8C in
various solvents
Product
Alcohol
2
8a
8b
8c
9a
9b
<0.1%a
1.8 ꢃ 0.2b
2.5 ꢃ 0.3c
0.7 ꢃ 0.1a
4.2 ꢃ 0.6b
7.0 ꢃ 0.8c
13.5 ꢃ 0.6d
2.6 ꢃ 0.3a
3.3 ꢃ 0.4b
2.9 ꢃ 0.2c
29 ꢃ 3a,d
26 ꢃ 1b,e
32 ꢃ 2c,f
8.3 ꢃ 1.2c
8.0 ꢃ 0.7j
ꢀ
ꢀ
ꢀ
Acyl Carbonate
10.8 ꢃ 1.6c
14.4 ꢃ 2.2c
2.7 ꢃ 0.4c
ꢀ
22 ꢃ 3j
ꢀ
ꢀ
ꢀ
29 ꢃ 3a,m
ꢀ
ꢀ
31 ꢃ 2a,p
ꢀ
Ester (Cage II)
Acid (Cage II)
11.7 ꢃ 0.5j
ꢀ
ꢀ
Scavenging
ꢀ
25 ꢃ 3a,d
25 ꢃ 2a,d
8.8 ꢃ 0.3a,n
22 ꢃ 2a,r
12.3 ꢃ 1.8c,i
18 ꢃ 1h,g
12.2 ꢃ 1.6c,t
28 ꢃ 2b,g
24 ꢃ 1c,g
Combination (Cage III)
22 ꢃ 1k,j
14.3 ꢃ 0.1l,j
16.0 ꢃ 0.1o
1.6 ꢃ 0.1a,s
a HCA.
b n-Octane or n-dodecane.
c C6D6.
d The average of yields at 70, 80, 90, 100, 108, and 123 8C in n-octane. Undecyl chloride.
e n-Undecane.
f n-Undecane and n-undecylbenzene.
g n-Docosane, determined by GC.
h n-Hexadecane, determined by GC in n-pentane.
i n-Octyl benzene-d5.
j Determined by GC in n-hexane.
k 1,4-Dicyclopentylbutane.
l 1,4-Diphenylbutane.
m Undecyl acetate.
n Methyl chloride. Thermolysis done in open tube, yield is lowered by the escape from the solution.
o n-Dodecane.
p Undecyl m-chlorobenzoate.
r m-Dichlorobenzene.
s 1-Chloro-3-undecylbenzene.
t n-Undecyl benzene-d5.
the singlet–triplet interconversion rate of radical pairs depends
on the sign and magnitude of the hyperfine coupling and
the g-factor difference between the paired radicals, the Boltzman
populations of the nuclear spin levels of the cage and escape
products are altered leading to enhanced product NMR signals
(anywhere between 10- and 500-fold larger that the signals in the
spectra of un-polarized products in the reaction mixture recorded
at temperature of thermolysis). Thus, the sign (E for emission
signal and A of the absorption signal), and the magnitude of
the polarization depend on the hyperfine coupling constants of
the nuclei coupled to the electron spins, differences between
the isotropic g-factors of the paired radicals (if any), radical
re-orientation and diffusion rates, and, in diacyl peroxides, the
lifetimes of acyloxy radicals.[28,39–41]Two types of enhancement
effects are seen:
Table 4. Yields of 1-undecene from 2, 9a, and 9b in different
solventsa
Solvent
2
9a
9b
n-C6H14
n-C8H18
iso-C8H18
ClC2H4Cl
C6H6
C6H5CH3
ꢀ
7.4 ꢃ 0.2
ꢀ
ꢀ
6.9 ꢃ 0.2
ꢀ
ꢀ
8.2 ꢃ 0.2
9.3 ꢃ 0.2
8.2 ꢃ 0.2
7.9 ꢃ 0.2
7.8 ꢃ 0.2
12.9 ꢃ 0.6
13.1 ꢃ 0.6
10.4 ꢃ 0.4
11.2 ꢃ 0.4
12.8 ꢃ 0.6
6.1 ꢃ 0.3
6.5 ꢃ 0.3
6.2 ꢃ 0.2
5.6 ꢃ 0.4
b
HCA, 14O 8C
HCA, hexachloroacetone.
a 80 8C, [peroxide]0 ¼ 10ꢁ4 M (unless noted otherwise).
Averages of at least of 2–4 samples per run.
b 0.1 M peroxide.
1. When the radicals within the pair possess different g-factors, as
do the alkyl and acyloxy radicals in Cage II, the result is a net
enhancement or the so-called ‘‘net effect’’ manifesting as a
wileyonlinelibrary.com/journal/poc
Copyright ß 2011 John Wiley & Sons, Ltd.
J. Phys. Org. Chem. 2011, 24 809–820

HIGH DECARBOXYLATION RATE OF AN ACYLOXY RADICAL
uniform enhancement of all the lines of the multiplet signals of
the cage and escape products provided the acyloxy radical
lives long enough for the polarization to built up. This can
be clearly seen from the general appearance of the 1H-CIDNP
spectra of the low and high field regions of decomposing 2,
9a, and 9b in HCA at 140 8C in Figs. 1 and 2. The lauroyl
m-chlorobenzoyl peroxide 9b with the slowest decarboxyla-
tion rate of the acyloxy radical 1a (m-chlorobenzoyloxyl,
upper traces) shows exclusively net enhancements. Note that
lines within every multiplet signal appear either all in emission
or all in absorption mode.
triplet. This is the result of the multiplet being a triplet (odd #
of lines) – multiplet effect signals with odd lines show zero
enhancement for the center line. Another interesting feature
of the 1H-CIDNP of decomposing 2 is the enhancement of the
protons of the combination product, n-dodecane. The E/A
polarized terminal methyl triplet of n-dodecane at 0.95 ppm
alters the signal of the terminal triplet of all product undecyl
chains such that the outer lines of the latter are partially erased
(low-filed line) and enhanced (high-field line).
It is of course possible for the acyloxy radical to live long
enough such that the polarization of the reaction products
exhibits the sum of polarizations from Cages II (net) and III
(multiplet) – the so-called ‘‘memory effect.’’ This is clearly the case
in the CIDNP spectra of 9a (middle traces). In this case, the
acetoxyl radical 1b, the acyloxy radical partner in Cage II,
decarboxylates just fast enough to allow for the multiplet effect to
be mixed into the net enhancements of undecyl chloride and
the n-undecyl acetate ester. From the average ratio of the CIDNP
intensity of the downfield to upfield signals of the terminal
protons of n-undecyl acetate produced in thermolysis of 9b, the
ratio of the net to multiplet effect is estimated to be 3.3 ꢃ 0.2,
that is, roughly 70% net and 30% multiplet effect. Note also
a significant absorption signal from methane, a Cage III
2. If the radicals within a pair are identical, or possess no
appreciable g-factor difference, as do the purely carbon-
centered radicals in Cage III, the result is a differential line
enhancement or the so-called ‘‘multiplet effect’’ manifesting as
E/A enhancement of different lines within the multiplet signals.
This is clearly the case in thermolysis of 2 (bottom traces in
Figs. 1 and 2), where all the polarized signals exhibit alternat-
ing enhancements of their multiplet lines. Since the lifetime
of the lauroyloxyl radical 1d in Cage II is extremely short,
only the multiplet enhancements from Cage III are observed.
1
The multiplet effect in the H-CIDNP spectrum from decom-
posing 2 is most pronounced for the a protons of undecyl
chloride at 3.5 ppm, which appear as a doublet instead of a
Figure 1. High-field region of the progressive-saturation 1H-CIDNP spectra from 9b (top), 9a (middle), and 2 (bottom) observed from 0.1 M peroxides
decomposing at 140 8C in HCA
J. Phys. Org. Chem. 2011, 24 809–820
Copyright ß 2011 John Wiley & Sons, Ltd.
wileyonlinelibrary.com/journal/poc

A. FRAIND ET AL.
Figure 2. Low-field region of the progressive-saturation 1H-CIDNP spectra from 9b (top), 9a (middle), and 2 (bottom) observed from 0.1 M peroxides
decomposing at 140 8C in HCA
disproportionation product from 9a. Thus, the spectra
1
and III – the separation of radical fragments prior to the
˚
presented here clearly show that the H-CIDNP is very sensitive
decomposition of the molecule, 2.6 ꢃ 0.6 A, and the reactive
˚
to the rate of acyloxy radical decarboxylation, even in mixed
diacyl peroxides giving rise to radical cages with different
acyloxy radicals.
encounter diameter, 2.1 ꢃ 0.4 A – were obtained from the
temperature/viscosity dependence of the probability of radical
re-encounter after the separation from the initial encounter at
time zero (b).[55–57] The average mass of n-undecyl and lauroyloxy
radicals was used, and the Stokes’ radius of both radicals was
assumed to be equal to the Stokes’ radius of the corresponding
Estimates of the decarboxylation rate of 1d from quantitative
studies of geminate products yields
[58,59]
˚
hydrocarbon undecane (2.4 A).
separation after the O–O bond scission was estimated to range
At 80 8C, the initial radical
In order to obtain the estimates of this rate from radical
cage product yields, we used the quantitative Kaptein–Den
Hollander radical-pair formulation[39–41,53,54] of CIDNP based on
Noyes model of radical dynamics in solution[55–57] to simulate the
temperature and viscosity dependence of the absolute yields,
using the decarboxylation rate constant as one of the adjustable
parameters. The variations of this approach are frequently used
to simulate radical product yields by neglecting any electron–
nuclear spin–spin interactions. Tables 5 and 6 show the detailed
GC analysis of the various thermolysis product yields from pero-
xide 2, at the very low (<10ꢁ4 M) initial concentrations.
For the purpose of data fitting, the radicals were assumed to
undergo a number of re-encounters after the initial O–O bond
homolysis, and the yields were calculated from the total number
of re-encounters. In this approach, the decarboxylation is treated
as a ‘‘reduction’’ of the re-encounter probabilities. The diffusion
parameters of the n-undecyl and lauroyloxyl radicals in Cages II
˚
˚
from 5.2 A (at lowest viscosity) to 4.4 A (at highest viscosity).
Using these parameters we obtain (0.9 ꢃ 0.5) ꢀ 1012 sꢁ1 for the
decarboxylation rate of lauroyloxyl radical 1d at 80 8C from the
temperature dependence of n-undecane and docosane
product yields from Cage III (Table 5). The constant temperature,
variable viscosity data from Table 6 yields 1.2 ꢃ 0.5 ꢀ 1012 sꢁ1
.
These values are certainly consistent with the absence of any net
CIDNP effects in thermolysis of 2, since they are three orders of
magnitude larger then that of acetyloxyl radical. The temperature
dependence of cage yields also allows for the qualitative estimate
of the activation energy for the lauroyloxyl radical decarboxyla-
tion. Even without numerical data processing, it is fairly apparent
from Table 6 that this activation energy is near zero – within the
error of the yield measurements there is very little to no variation
of the Cage II product yields with temperature. In fact, the plots of
the yields versus 1/T results in level straight lines. The same lack of
wileyonlinelibrary.com/journal/poc
Copyright ß 2011 John Wiley & Sons, Ltd.
J. Phys. Org. Chem. 2011, 24 809–820

HIGH DECARBOXYLATION RATE OF AN ACYLOXY RADICAL
Table 5. Viscosity dependence of Cage II and III product yields from 2a
h (cP)
1-Undecene cage yield
n-Undecane cage yield
Excess 1-undecene cage yield
n-Docosane cage yield
0.134b
0.178c
0.210d
0.271e
0.28f
0.0585 ꢃ 0.0027
0.0205 ꢃ 0.0009
0.0380 ꢃ 0.0028
0.177 ꢃ 0.012
0.208 ꢃ 0.015
0.223 ꢃ 0.013
0.260 ꢃ 0.015
0.232 ꢃ 0.014
0.276 ꢃ 0.015
0.310 ꢃ 0.024
0.0636 ꢃ 0.0023
0.0226 ꢃ 0.0013
0.0410 ꢃ 0.0024
0.0675 ꢃ 0.0018
0.0256 ꢃ 0.0011
0.0419 ꢃ 0.0019
0.0685 ꢃ 0.0019
0.0286 ꢃ 0.0010
0.0399 ꢃ 0.0020
ꢀ
ꢀ
ꢀ
ꢀ
ꢀ
ꢀ
ꢀ
ꢀ
ꢀ
0.361g
0.429h
a Note that here the actual relative yields are shown (not % yields as before), temperature 80 8C, initial peroxide concentration
1 ꢀ 10ꢁ4 M.
b n-Pentane.
c n-Pentane (mole fraction 0.68) and iso-octane.
d n-Pentane (mole fraction 0.42) and iso-octane.
e n-Octane.
f iso-Octane.
g n-Nonane.
h n-Decane.
temperature variation is observed for same Cage II products
from peroxides 8a–c, Table 7. See Supplementary information for
the equations used for the calculation of product yields and
loxy[36–38] radicals. With very few exceptions, these studies used
a semi-quantitative approach that requires independent esti-
mates of the relaxation times of products, whose polarization
is observed and simulated numerically. This would not apply
here, since the relaxation times of key protons in this study
vary from 1.5 s for the a protons of n-undecyl chloride at 25 8C
to 41.5 s for the internal proton of the double bond in 1-undecene
at 140 8C (determined at 400 MHz by the inversion-recovery
method).[39] Additionally, many CIDNP and yields analysis
studies are complicated by the fact that either the reactive
solvents (methanol or acetone) were used,[36–38] or the
solvent did not have sufficient scavenging abilities (as in the
case of CCl4). Therefore, we utilized the quantitative saturation-
recovery[45] CIDNP acquisition techniques (see Experimental
1
quantitative H-CIDNP enhancement factors.
Estimates of the decarboxylation rate of 1d from
1
quantitative H-CIDNP studies
1H-CIDNP spectroscopy can, in principle, also be used to
quantitatively estimate the decarboxylation rates of acyloxy
radicals 7 as a crosscheck on lifetime estimates obtained
from cage product yields. There has been a number of
previous studies that employed CIDNP spectroscopy to quantify
the lifetimes of benzoyloxy,[27,28] acetyloxyl[11] and propanoy-
Table 6. Temperature dependence of Cage II and III product yields from 2a
T (8C)
h (cP)
T/h
1-Undecene cage yield
n-Undecane cage yield
Excess 1-undecene cage yield
80
85
96
105
110
115
0.318
0.304
0.278
0.258
0.249
0.240
1110.5
1178.1
1327.9
1465.7
1538.8
1617.3
0.0630 ꢃ 0.0018
0.0618 ꢃ 0.0020
0.0585 ꢃ 0.0020
0.0575 ꢃ 0.0022
0.0559 ꢃ 0.0016
0.0554 ꢃ 0.0016
0.0294 ꢃ 0.0010
0.0289 ꢃ 0.0010
0.0280 ꢃ 0.0012
0.0275 ꢃ 0.0010
0.0272 ꢃ 0.0010
0.0269 ꢃ 0.0010
0.0336 ꢃ 0.0021
0.0329 ꢃ 0.0022
0.0305 ꢃ 0.0023
0.0300 ꢃ 0.0024
0.0287 ꢃ 0.0019
0.0285 ꢃ 0.0019
T (8C)
h (cP)
T/h
1-Undecene cage yield
n-Docosane cage yield
n-Undecyl lauroate cage yieldb
70
80
90
100
108
123
0.312
0.283
0.259
0.238
0.220
0.204
1099.8
1247.9
1402.1
1567.9
1732.5
1941.9
ꢀ
ꢀ
ꢀ
ꢀ
ꢀ
ꢀ
0.235 ꢃ 0.012
0.243 ꢃ 0.013
0.256 ꢃ 0.013
0.254 ꢃ 0.013
0.262 ꢃ 0.013
0.262 ꢃ 0.014
0.131 ꢃ 0.011
0.133 ꢃ 0.011
0.138 ꢃ 0.014
0.140 ꢃ 0.012
0.133 ꢃ 0.012
0.141 ꢃ 0.013
a Note that here the actual relative yields are shown (not % yields), initial peroxide concentration 1 ꢀ 10ꢁ4 M in toluene.
b Initial peroxide concentration 0.1 M in n-octane.
J. Phys. Org. Chem. 2011, 24 809–820
Copyright ß 2011 John Wiley & Sons, Ltd.
wileyonlinelibrary.com/journal/poc

A. FRAIND ET AL.
Table 7. Temperature dependence of products yields from 8a–ca,b
T, 8C
Alkene
Alkane
Excess alkenec
Ester
Combination Cage III
8a
80d
110
8b
80e
100
6.0 ꢃ 0.3
5.2 ꢃ 0.1
3.5 ꢃ 0.1
2.7 ꢃ 0.1
2.5 ꢃ 0.1
2.5 ꢃ 0.1
ꢀ
ꢀ
ꢀ
ꢀ
7.4 ꢃ 0.2
6.3 ꢃ 0.1
5.4 ꢃ 0.1
2.4 ꢃ 0.1
2.0 ꢃ 0.1
1.8 ꢃ 0.1
5.0 ꢃ 0.2
4.3 ꢃ 0.1
3.6 ꢃ 0.1
21.7 ꢃ 1.4
20.9 ꢃ 0.4
18.7 ꢃ 0.5
21.2 ꢃ 0.4
20.8 ꢃ 0.6
19.7 ꢃ 0.7
120
8c
80
100
120
8.0 ꢃ 0.3
7.7 ꢃ 0.3
7.4 ꢃ 0.5
2.9 ꢃ 0.1
3.0 ꢃ 0.1
2.6 ꢃ 0.1
5.1 ꢃ 0.3
4.7 ꢃ 0.3
4.8 ꢃ 0.4
11.7 ꢃ 0.5f
14.0 ꢃ 2.3g
12.2 ꢃ 1.9h
14.3 ꢃ 0.2h
15.7 ꢃ 1.9g
14.6 ꢃ 1.4h
a Yields relative to 1 mol of decomposed peroxide; errors represent 1 SD.
b Alkene, alkane, ester, combination are – 8a: 1-octene, n-octane, n-octyl nanoate, n-hexadecane; 8b: vinylcyclopentane, ethyl-
cyclopentane, 2-cyclopentylethyl 3-cyclopentylpropionate, 1,4-dicyclopentylbutane; 8c: styrene, ethylbenzene, 2-phenylethyl
3-phenylpropionate, 1,4-diphenylbutane.
c Yield of alkene ꢁ yield of alkane.
d Six different samples analyzed using one ratio of int. strd. to 8a. 1 ꢀ 10ꢁ4 M [8a]o in n-pentane.
e Ten different samples analyzed using two different ratios of int. strd. to 8b. 2 ꢀ 10ꢁ4 and 4 ꢀ 10ꢁ4 M [8b]o in n-pentane.
f Four different samples analyzed using two different ratios of int. strd. to 8c. 2 ꢀ 10ꢁ4 M [8c]o in n-hexane.
g Nine different samples analyzed using two different ratios of int. strd. to 8c. 2 ꢀ 10ꢁ4 M [8c]o in n-hexane.
h Sixteen different samples analyzed using one ratio of int. strd. to 8c. 2 ꢀ 10ꢁ4 M [8c]o in n-hexane.
Section) that eliminate distortions due to the relaxation time
differences. All CIDNP spectra were acquired in HCA to ensure
complete scavenging of escaped radicals.[14]
an adjustable value. HCA viscosity was approximated by that of
n-dodecane at 140 8C (0.024 poise).[58] The absolute enhance-
ment factors were determined as the ratio of the population
difference between the spin states for individual product proton
levels involved in the transition (actual line intensity), and the
intensity of the corresponding line in the product signal due
solely to thermal polarization (at 140 8C, the temperature of the
CIDNP spectra acquisition).[39–42] Since the CIDNP spectra often
display considerable polarization even for very minor products,
the enhancement factors were corrected for the number of scans
required to obtain the spectra of the decomposition products.
The observed and calculated signal polarizations of thermo-
lysis products of 2 and 9a are shown in Table 8. This data
The quantitative model of the radical dynamics in solution
discussed above was adapted to CIDNP calculations through the
introduction of spin-dependent terms involved in the buildup of
NMR signal polarization. These included the isotropic g-factors for
the n-undecyl and lauroyloxyl radicals in Cages II and III, and
isotropic hyperfine coupling constants to a and b protons in the
n-undecyl radical, required for the calculation of the S–T0 mixing
rates.[28,39–42] Other model parameters (diffusion coefficients,
radical radii, etc.) were as described above, and the decarboxyla-
tion rate of the acyloxy radical in Cage II was once again used as
1
Table 8. Yields and H enhancement factors in thermolysis of 2 and 9aa
1-Undecene
Ester
n-Dodecane
1-Undecyl chloride
2
Yields (calc.)
Observed
Enhancement factors (calc.)
Observed
9a
Yields (calc.)
Observed
Enhancement factors (calc.)
Observed
2.5
3.5 ꢃ 0.5
20.0
11.3 ꢃ 1.2
ꢁ200
ꢀ
ꢀ
26.3
28.8 ꢃ 3.2
þ89
ꢁ50
þ385 (n-docosane)
Not resolved
ꢁ35 ꢃ 4
None
þ99 ꢃ 11
8.9
7.8 ꢃ 0.2
33.0
29.8 ꢃ 2.6
ꢁ22
5.5
16.0 ꢃ 0.1
26.3
25.1 ꢃ 0.3
ꢁ60
þ385
þ128
ꢁ71 ꢃ 18
ꢁ17 ꢃ 4
þ200 ꢃ 41
þ239 ꢃ 49
a [peroxide]0 ¼ 0.1 M at 140 8C. Averages of at least of 4–8 samples per run. Enhancement factors are for the a-protons in ester and
undecyl chloride, terminal alkene protons in 1-undecene, and interior protons in n-dodecane.
wileyonlinelibrary.com/journal/poc
Copyright ß 2011 John Wiley & Sons, Ltd.
J. Phys. Org. Chem. 2011, 24 809–820

HIGH DECARBOXYLATION RATE OF AN ACYLOXY RADICAL
represents the best overall agreement between the cal-
culated yields and enhancement factors and those observed
experimentally. Here we used the decarboxylation rates of
0.5 ꢀ 1012 sꢁ1 for the lauroyloxyl radical 1d, and 4 ꢀ 109 sꢁ1 for
the acetyloxyl radical 1b (at 140 8C in good agreement with the
literature values for the latter).[28] Both are consistent with
those predicted from temperature and viscosity dependence
of the absolute radical product yields discussed above. Note
however, two very important points: (A) these decarboxylation
rate values fail to correctly predict the observed enhancement
factors for the ester signal in thermolysis of 2. As seen in Fig. 2,
the CIDNP spectra show no polarized ester signal; and (B) the
polarization observed for the internal and terminal protons in
1-undecene is built up in the reactive and diffusional processes in
Cage III, not Cage II indicating that the excess alkene likewise
shows no polarized signal.
(0.7–5.0) ꢀ 1011
s ,
ꢁ1 [20,22] or the decarboxylation of the benzilate
anion-derived Ar2C(OH)CO2ꢂ (8.0 ꢀ 1011
s
ꢁ1).[35]
As was reported previously (vide supra), no net effects
were detected in the quantitative CIDNP spectra of decomposing
lauroyl peroxide 2. It therefore appears that net CIDNP enhance-
ments are only observable for acyloxy radicals that decarboxylate
with the rate of at most 109 sꢁ1. A rate two to four times this
number, especially an order of magnitude larger, erases all net
CIDNP effects, yet appears to still give rise to detectable
yields of the combination and disproportionation products of
acyloxy-alkyl radical pair. This can be explained by the fact
that the vibrationally ‘‘hot’’ alkyl and acyloxy radicals react within
a ‘‘primary’’ Cage II in very few, if any, diffusional steps. In fact, the
combination and disproportionation of these radicals is likely
regulated by their spatial re-orientation, and simple mechanical
displacement of forming carbon dioxide, within a ‘‘primary cage’’
rather than diffusion and subsequent ‘‘secondary cage’’ reactions.
By contrast, the radicals in Cage III have ample opportunity
to diffuse and equilibrate thermally during their polarization and
reactions. As the result, calculation of the yields and EF based on
the standard diffusion CIDNP model does not apply to Cage II
products. This is particularly evident from Table 5, where no
viscosity effect is seen on the yield of excess 1-undecene.
At present, no model is available to simulate induced polarization
in ultra-fast spin-selective primary radical cage reactions; the
latest approaches using Monte Carlo simulations of diffusional
trajectories,[60] or the backward stochastic Liouville equation[61]
also assume S–T0 mixing of freely diffusing radicals.
Finally, it is desirable to correlate the findings presented here
with the current theoretical ideas about the electronic structure,
and potential energy surfaces of acyloxy radicals. These have
been investigated computationally for the flouro, formyloxyl,
acetoxyl, and benzoyloxy radicals using techniques ranging
from CAS-SCF and CASPT2,[62,63] to DFT-B1PW91 and B1LYP,[64]
and coupled cluster theory based computations,[65] experimen-
tally using dissociative photodetachment,[66] low-temperature
anion photoelectron spectroscopy,[67] direct EPR detection,[68]
and high resolution rotational MMW (millimeter wave) spec-
troscopy.[69] For the formyloxyl radical, the experiment[68,69] and
DISCUSSION
We believe that the data presented here provides accurate
estimates for the decarboxylation rate of lauroyloxyl radical, 1d
because it eliminates two potential arguments against the
existence of 1d as an intermediate in thermal decomposition of
lauroyl peroxide 2. First is the interference from the polar
decomposition pathway. The presence of small amounts of ester
and carboxylic acids in thermolysis of primary aliphatic diacyl
peroxides is commonly ascribed to polar decomposition
mechanisms.[46,47,49,50] However, we feel that for the peroxides
discussed here, the polar pathway to ester and excess alkene can
be safely eliminated. Our previous studies with threo-2,3-d2
peroxide
2 showed complete deuterium scrambling in
1-undecene, indicating a purely radical source.[11] Additionally,
a very similar 4,4-dimethylpentanoyl m-chlorobenzoyl peroxide
and its a- and b-dideuteriated analogues, peroxides highly
susceptible to the intervention of the ionic pathways and polar
rearrangement of the resulting 3,3-dimethylbutyl cation, gen-
erate nearly identical (13–16%) yields of the un-rearranged ester,
3,3-dimethylbutyl m-chlorobenzoate, exclusively from radical
reaction pathway.[46] Finally, as illustrated by the data in
Table 4, the yields of the products ascribed to the reactions of
the acyloxy-alkyl radical pairs in thermolysis of 2 and 9a and 9b
(potentially highly capable of polar rearrangements) are
independent of solvent polarity.
2
calculations suggest a A1 (C2v) ground state, with the low-lying
2B1 state 4 kJ/mol higher in energy. The other important low
2
energy state is A0 (Cs); it is a transition state for the HCOO to
HOCO isomerization.[63] The acetyloxyl radical presents a more
complicated case. There are two Cs species differing in the relative
orientation of one of the methyl group’s CH bond, that is, in
the OCO plane or perpendicular to it.[62–64] The two sates of the
former are equivalent to 2A1 and 2B1 states of the formyloxyl
Another objection to the existence of acyloxy radical inter-
mediates in thermal decomposition of primary diacyl peroxides
was the potential for excess alkene and ester products resulting
from bimolecular reactions between the peroxide and un-
scavenged escaped radicals. In this study, we examined the
product yields under the conditions of complete oxygen
scavenging of escaped radicals at very low peroxide concen-
trations, eliminating the side reactions between the peroxide
and its radical products. Therefore, we are highly confident that
the qualitative and quantitative analysis of ester and excess
alkene yields in decomposition of 2, 8a–c, and 9a,b strongly
indicates the presence of a short-lived acyloxyl-alkyl radical pairs
in all these symmetrical and mixed diacyl peroxides. Accordingly,
we are also confident in our estimate of the lauroyloxyl radical
2
radical, while the lowest state of the latter is A00(B2). With
identical C–O bond lengths, delocalized spin density, and free
rotation of the methyl group it is the lowest in energy of all four
states. The next energy state, A0(A1) is some 50 kJ/mol higher,
2
˚
with an elongated 1.68 A C–C bond. It is the dissociative state to
CH3 radical and CO2. The instability of this state is due to the
increased steric repulsion between the CH3 and CO2 subunits
resulting from a compressed 106.68 H–C–O bond angle, and
highly open 143.68 O–C–O angle (much larger than 110.28 in the
ground 2A00(B2) state, and 110.38 in 2A0(B2)).[62–64] Any rotation of
the CH3 group in this state leads to dissociation. Thus, the
much greater decarboxylation rate constant for the long chain
radicals such as lauroyloxyl (1d) is likely a steric effect of a very
large substituent greatly destabilizing the dissociative 2A0(A1)-like
electronic states.
decarboxylation rate constant of (0.5–1.5) ꢀ 1012
s
ꢁ1. This makes
it one of the fastest decarboxylating radicals, faster in fact than
the very rapid ring-opening of the cyclopropylcarbinyl radical
J. Phys. Org. Chem. 2011, 24 809–820
Copyright ß 2011 John Wiley & Sons, Ltd.
wileyonlinelibrary.com/journal/poc

A. FRAIND ET AL.
[31] B. Abel, J. Assmann, M. Buback, C. Grimm, M. Kling, S. Schmatz, J.
Schroeder, T. Witte, J. Phys. Chem. A 2003, 107, 9499.
[32] J. Aschenbrucker, M. Buback, N. P. Ernsting, J. Schroeder, U. Steeg-
muller, J. Phys. Chem. B 1998, 102, 5552.
[33] D. E. Falvey, G. B. Schuster, J. Am. Chem. Soc. 1986, 108, 7419.
[34] B. Abel, M. Buback, M. Kling, S. Schmatz, J. Schroeder, U. Steegmuller,
J. Am. Chem. Soc. 2003, 125, 13274.
[35] T. M. Bockman, S. M. Hubig, J. K. Kochi, J. Org. Chem. 1997, 62, 2210.
[36] E. D. Skakovskii, A. I. Stankevich, L. Tychinskaya, O. Shirokii, Yu.
Choban, S. V. Rykov, Russ. J. Gen. Chem. 2004, 74, 1719.
[37] E. D. Skakovskii, A. I. Stankevich, L. Yu. Lamotkin, S. V. Rykov, Russ.
J. Gen. Chem. 2001, 71, 614.
Acknowledgements
Part of this research was performed while LR was a graduate
student at Brown University, Providence, RI. The author is greatly
indebted to Prof. Ronald G. Lawler for suggesting this project and
providing lab space, encouragement, and guidance during the
authors stay at Brown University, and during the preparation of
this manuscript. AF and LR both acknowledge the support Office
of University Research Services at Towson, and AF is very grateful
to the Raspet family for the support from the Raspet Summer
Fellowship Endowment.
[38] E. D. Skakovskii, A. I. Stankevich, L. Tychinskaya, L. Yu. Lamotkin, S. V.
Rykov, Russ. J. Gen. Chem. 1997, 68, 1028.
[39] L. R. Ryzhkov, Ph.D. Thesis, Brown University, Providence, RI, USA, 1991.
[40] J. Den Hollander, Chem. Phys. Lett. 1975, 10, 167.
[41] R. Kaptein, Adv. Free-Radic. Chem. 1975, 4, 319 and references
therein.
REFERENCES
[42] H. Hayashi, Introduction to Dynamic Spin Chemistry. Magnetic Field
Effects on Chemical and Biochemical Reactions, Vol. 8, World Scientific
Publishing Co., Hackensack, NJ. 2004.
[43] R. A. Cooper, R. G. Lawler, H. R. Ward, J. Am. Chem. Soc. 1972, 94,
545.
[44] W. H. Urry, M. S. Kharasch, J. Am. Chem. Soc. 1944, 66, 1438.
[45] R. G. Lawler, P. F. Barber, J. Magn. Res. 1980, 40, 135.
[46] S.-G. Lee, J. Chem. Soc. Perkin Trans. 1993, 2, 1361.
[47] Y.-C. Liu, Q.-X. G, Acta Chim. Sinica 1985, 43, 1172.
[48] J. Guillet, J. Gilmer, Can. J. Chem. 1969, 47, 4405.
[49] C. Walling, H. Waits, J. Milanovic, C. Pappiaonnou, J. Am. Chem. Soc.
1970, 92, 4928.
[1] D. H. R. Barton, P. Blundell, J. Cs. Jaszberenyi, J. Am. Chem. Soc. 1991,
113, 6937.
[2] D. A. Whiting, S. P. Green, J. Chem. Soc. Chem. Commun. 1994, (21),
2441.
[3] F. E. Ziegler, P. G. Harran, Tetrahedron Lett. 1993, 34, 4505–4508.
[4] F. E. Ziegler, P. G. Harran, J. Org. Chem. 1993, 58, 2768.
[5] K. Fujimory, Organic Peroxides, Wiley, NY, 1992.
[6] M. Newcomb, S.-Y. Choi, J. H. Horner, J. Am. Chem. Soc. 1999, 64, 1225.
[7] M. Newcomb, Tetrahedron 1993, 49, 1151.
[8] M. Newcomb, P. H. Toy, Acc. Chem. Res. 2000, 33, 449.
[9] M. Newcomb, M.-H. Le Tadic-Biadatti, D. L. Chestney, E. S. Roberts, P. F.
Hollenberg, J. Am. Chem. Soc. 1995, 117, 12085.
[10] A. Fraind, R. Turncliff, T. Fox, B. Salisbury, J. Sodano, A. Elabd, L. R.
Ryzhkov, J. Undergrad. Chem. Res. 2007, 6, 67.
[50] J. Smid, A. Rembaum, M. Szwarc, J. Am. Chem. Soc. 1956, 78,
3315.
[51] K. M. Salikhov, Yu. N. Molin, R. Z. Sagdeev, A. L. Buchachenko, Spin
Polarization and Chemical Effects in Radical Reactions, Elsevier,
Amsterdam, 1984.
[11] L. R. Ryzhkov, J. Org. Chem. 1996, 61, 2801.
[12] K. Fujimore, Y Oshibe, Y Hirose, S. Oea, J. Chem. Soc. Perkin Trans.
1996, 2, 405.
[52] L. T. Muus, P. W. Atkins, K. A. McLauchlan, J. B. Pedersen, Chemically
Induced Magnetic Polarization, D. Reidel, Dordrecht, 1977.
[53] J. A. Den Hollander, Ph.D. Thesis; University of Leiden, Leiden, Nether-
lands, 1976.
[13] K. Fujimore, Y Oshibe, Y Hirose, S. Oea, J. Chem. Soc. Perkin Trans.
1996, 2, 413.
[14] R. Hiatt, Organic Peroxides, Vol. II, Wiley-Interscience, NY, 1971.
[15] S. F. K. Oae, The Chemistry of Peroxides, Wiley, New York, 1983, Ch. 18.
[16] T. Koenig, Free Radicals, Vol. I, Wiley-Interscience, NY, 1973.
[17] P. S. Skell, D. D. May, J. Am. Chem. Soc. 1981, 103, 967.
[18] Z. Zhou, X. Cheng, X. Zhou, H. Fu, Chem. Phys. Lett. 2002, 353, 281.
[19] M. M. Maricq, J. J. Szente, J. S. Francisco, J. Phys. Chem. 1996, 100,
4514.
[20] S. Y. Choi, M. Newcomb, Tetrahedron 1995, 114, 10915.
[21] M. Newcomb, T. R. Varick, C. C. Johnson, M. B. Manek, J. Am. Chem.
Soc. 1992, 114, 10915.
[22] M. Newcomb, A. A. Martin-Esker, C. C. Johnson, J. H. Horner, J. Am.
Chem. Soc. 1994, 116, 9174.
[23] E. G. Janzen, C. A. Evans, T. Nishi, J. Am. Chem. Soc. 1972, 104, 4515.
[24] J. Chateauneuf, J. Lusztyk, K. U. Ingold, J. Am. Chem. Soc. 1988, 110,
2877.
[54] J. Deutsch, J. Chem. Phys. 1960, 56, 6076.
[55] R. Noyes, D. Booth, J. Am. Chem. Soc 1960, 82, 1868.
[56] R. Noyes, J. Am. Chem. Soc. 1956, 78, 5486.
[57] R. Noyes, J. Chem. Phys. 1954, 22, 1349.
[58] H. Fischer (Ed.), Landoldt-Bornstein (LB Series) Compilation of Physical
Data, Vol. 5a, Series II, Springer-Verlag, Berlin, 1977.
[59] D. Douglas, D. Mccall, J. Chem. Phys. 1958, 62, 1102.
[60] M. Goez, R. Heun, J. Phys. Chem. A 2001, 105, 10446.
[61] J. B. Pedersen, M. Christensen, J. Phys. Chem. B 2004, 108, 9516.
[62] A. Rauk, P. Borowski, B. Roos, Chem. Phys. 1995, 197, 73.
[63] A. Rauk, D. Yu, J. Armstrong, J. Am. Chem. Soc. 1994, 116, 8222.
[64] M. Kieninger, O. N. Ventura, S. Suhai, Int. J. Quant. Chem. 1998, 70,
253.
[65] J. Breidung, W. Thiel, J. Phys. Chem. A 2006, 110, 1575.
[66] Z. Lu, R. E. Continetti, J. Phys. Chem. A 2004, 108, 9962.
[67] X.-B. Wang, H.-K. Woo, L.-S. Wang, B. Minofar, P. Jungwirth, J. Phys.
Chem. A 2006, 110, 5047.
[25] K. Ingold, J. Lusztyk, J. Chateanneuf, J. Am. Chem. Soc. 1988, 110,
2886.
[26] S. Yamauchi, N. Hirota, S. Takahara, H. Sakuragi, K. Tokamaru, J. Am.
Chem. Soc. 1985, 107, 5021.
[68] H. Beckers, H. Willner, D. Grote, W. Sander, J. Geier, J. Chem. Phys.
2008, 128, 224301.
[69] L. Kolesnikova, J. Varga, H. Beckers, M. Simeksova, Z. Zelinger, L. Nova
Striteska, P. Kania, H. Willner, S. Urban, J. Chem. Phys. 2008, 128,
224302.
[27] R. Schwerzel, R. Lawler, G. Evans, Chem. Phys. Lett. 1974, 29, 106.
[28] R. Kaptein, J. Brokken-Zijp, F. de Kanter, J. Am. Chem. Soc. 1972, 94,
6280.
[29] J. A. Pincock, Acc. Chem. Res. 1997, 30, 43.
[30] J. W. Hilborn, J. A. Pincock, J. Am. Chem. Soc. 1991, 113, 2683.
wileyonlinelibrary.com/journal/poc
Copyright ß 2011 John Wiley & Sons, Ltd.
J. Phys. Org. Chem. 2011, 24 809–820